The Proposition of a New Multi-Agent Access Mechanism Based on the Fairness Principle
No Thumbnail Available
Date
2022
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
A multi-agent system can be defined as a set of entities able to interact with each
other and the environment to execute complex tasks. Usually, this kind of system is
composed of reactive agents playing different roles. The intelligence of the system emerges
from the behaviors undertaken by the agents. These behaviors can be categorized into two
types, collaborative and competitive behaviors.
Multi-agent Systems (MAS) often must pursue a common goal to realize that they
have to form an efficient coalition. The Pursuit-Evasion game (PE) is a MAS np-hard
problem in which the pursuers' agents must block the evaders' movement while considering
the temporal constraint. On the one hand, this problem can be processed via the introduction
of a coalition formation algorithm allowing the pursuers to be grouped in different pursuit
groups according to the detected evaders in the environment. On the other hand, it can be
treated via path planning approaches that permit the displacements of the pursuers.
An approach for coalition formation of multi-agent fairness-based and Agent Group
Role Membership Function-based model is proposed, thus making a unique algorithm
Fairness-based AGRMF known as FBAGRMF. To achieve this, we have completely
changed the organization method and created a group access mechanism that creates
balanced pursuit groups and compared it to the existing approach of coalition formation in
an existing PE. The simulation results reflect this new algorithm's advantages and
collaboration between agents in MAS.
Keywords: Multi-agent systems; Organization; Coalition formation; Pursuit-Evasion;
Markov Decision Process.